Vertebrate cone and rod photoreceptor cells use similar mechanisms to transduce light signals into electrical signals, but their responses to light differ in sensitivity and kinetics. To assess the role of G-protein GTP hydrolysis kinetics in mammalian cone photoresponses, we have characterized photoresponses and GTPase regulatory components of cones and rods from the cone-dominant retina of the eastern chipmunk. Sensitivity, based on the stimulus strength required for a half-maximum response, of the M-cone population was 38-fold lower than that of the rods. The relatively lower cone sensitivity could be attributed in part to lower amplification in the rising phase and in part to faster recovery kinetics. At a molecular level, cloning of chipmunk cDNA and expression of recombinant proteins provided standards for quantitative immunoblot analysis of proteins involved in GTPase acceleration. The ratio of the cGMP-phosphodiesterase inhibitory subunit γ to cone pigment, 1:68, was similar to the levels observed for ratios to rhodopsin in bovine retina, 1:76, or mouse retina, 1:65. In contrast, the ratio to pigment of the GTPase-accelerating protein RGS9-1 was 1:62, more than 10 times higher than ratios observed in rod-dominant retinas. Immunoprecipitation experiments revealed that, in contrast to rods, RGS9-1 in chipmunk retina is associated with both the short and long isoforms of its partner subunit Gβ5. The much higher levels of the GTPase-accelerating protein complex in cones, compared with rods, suggest a role for GTPase acceleration in obtaining rapid photoresponse kinetics.
Humans and other diurnal animals rely primarily on their cone photoreceptors for functional vision. Daylight vision, color discrimination, rapid motion detection, and finely focused images all rely on cones (for review, see Masland, 2001). Humans lacking rod function but retaining cone function have the relatively mild condition of night blindness. In contrast, those retaining rod function but lacking cone responses have very severe visual defects, (for review, see Birch, 1999; Hicks and Sahel, 1999). Therefore, it is of great interest to understand the biochemical properties of cones and the relationships between their molecular and physiological properties. Cone photoresponses have been recorded from some mammalian species (Kraft, 1988; Schnapf et al., 1990; Kraft et al., 1998; Schneeweis and Schnapf, 1999), but biochemical data are, for the most part, lacking, because no procedures have been developed for isolation of cone outer segments in amounts comparable with those obtainable from rod outer segments. Consequently we have a wealth of biochemical information about rods but only rudimentary information on the biochemistry of mammalian cones. In rod cells, quantification of the amounts of phototransduction components and characterization of their biochemical properties allow detailed modeling of the molecular mechanisms of phototransduction (Detwiler et al., 2000; Hamer, 2000;Leskov et al., 2000; Arshavsky et al., 2002). In cones, similar biochemical components are present, but their concentrations are mostly unknown.
Retinas of cone-dominant animals represent useful starting material for biochemical analysis of cone phototransduction as well as for electrophysiological recording. Two readily available examples are ground squirrels (Kraft, 1988) and chipmunks. We have analyzed photoresponses of the eastern chipmunk Tamias striatus and measured amounts of proteins essential for their inactivation phases. Because these proteins are found almost exclusively in photoreceptor outer segments, and because rods make a negligible contribution to their amounts, this quantification can be performed on crude extracts of total retina without purification of cone outer segments. Because the detection method relies on recognition by antibodies raised against orthologous proteins from other species, it was necessary to clone cDNA encoding the chipmunk proteins and to express them for accurate calibration of antibody sensitivity.
Our focus has been on proteins that determine the lifetime of transducin activation through regulation of the kinetics of G-protein GTP hydrolysis. The reason is that major differences between cones and rods include much lower sensitivity and faster recovery kinetics in cones. Faster recovery may contribute to lowering sensitivity. In both cones and rods, recovery is the slow phase of photoresponses, and it has been proposed that the rate-limiting step in rod transduction is transducin GTP hydrolysis (Sagoo and Lagnado, 1997; Nikonov et al., 1998). Immunostaining in bovine and human retina (Cowan et al., 1998;Zhang et al., 1999) suggested that cones had higher RGS9-1 content than rods but did not allow quantitative estimation of relative amounts.
We report here results from electrophysiological recordings of light responses in chipmunk rods and cones, along with quantification of the GTPase accelerating proteins RGS9-1, Gβ5, and cGMP phosphodiesterase inhibitory subunit γ (PDEγ) in chipmunk retina.
Materials and Methods
Buffers. Standard buffers were buffer A (in mm: 10 3-(N-morpholino) propane sulfonic acid (MOPS), pH 7.0, 30 NaCl, 60 KCl, 2 MgCl2, and 1 DTT and ∼20 mg/l phenylmethylsulfonyl fluoride), buffer B (in mm: 25 Tris and 192 glycine, pH 8.3), buffer C (in mm: 10 MOPS, 1 MgCl2, 50 NaCl, and 0.1 EDTA, pH 8.0), and buffer D (in mm: 10 HEPES, 100 NaCl, and 2 MgCl2). Other buffer components and conditions were varied as indicated throughout.
Animal and tissue preparation. The eastern chipmunk is common in rural and suburban neighborhoods of the southeastern United States. Animals were trapped by permit in Jefferson County Alabama. Animals were housed singly with a hiding tube and ad libitumaccess to food and water; captivity was well tolerated, because some animals were kept for >4 years. All experimental protocols were approved by the University of Alabama Institutional Animal Care and Use Committee. To start, a chipmunk was dark-adapted overnight and then killed by carbon dioxide asphyxiation. Details of the tissue preparations and electrolytes have been given previously (Kraft et al., 1993). Briefly, the retina was isolated under infrared light in Lebovitz's L-15 medium and stored at 4°C in the dark. Experiments were performed on the same retina for 2–3 d; each tissue sample lasted 3–4 hr. For each experiment, one-third of a retina (∼3 × 4 mm) was removed from cold storage and chopped under infrared light to produce small pieces of retina ∼50–100 μm on a side and then warmed to near body temperature in a perfusion chamber. The circulating dark current of individual rods or cones was recorded by drawing the outer segment into a suction electrode, whose inner diameter matched the outer segment diameter of the cell. The photocurrent and stimulus monitor signals were digitized with hardware (MIO16) and software (LabView) from National Instruments (Austin, TX). A stimulus set consisted of 5–30 responses to the same wavelength and intensity of light. The light bench focused a 440-μm-diameter spot of light at the plane of the cells. The wavelength was controlled with three-cavity interference filters (Andover Corp., Salem, NH) with an average bandwidth of 10 nm. Neutral density filters (Reynard Corp., San Clemente, CA) attenuated the light. Calibration of unattenuated light at each wavelength was performed daily with a photometer (model 350; Graseby Optronics, Orlando, FL).
Measuring the action spectra. Spectral sensitivity of the visual pigments was estimated by measuring the action spectra using the criterion response method and the photocurrent responses of individual cones. The action spectrum was determined for up to 20 wavelengths at ∼20 nm intervals between 380 and 760 nm. Spectral sensitivity was measured by adjusting the intensity of light at each wavelength to produce a criterion response of ∼25% of the maximum current. The sensitivity at each wavelength was measured relative to a standard wavelength, (500 nm). Initially, for each cell, the complete intensity response function was determined at 500 nm. Subsequently, for each test wavelength, two or three light intensities were used to obtain current responses of 10–60% of maximum. Sensitivity measures at the standard wavelength were repeated after every two or three test wavelengths to avoid errors attributable to changes in the physiologic state of the cell or electrode seal.
Calculations of phototransduction gain. The gain factor for the rising phase of the photoreceptor response was estimated by fitting Equation 22 from Pugh and Lamb (1993), given below as Equation 1, to the initial portion of the rising phase of the photocurrent response. The fitting was performed simultaneously to responses to two to five intensities in the linear range (see Fig. 2 C,D). Equation 1where R max is the maximum response; t d is a combined delay factor for all the biochemical reactions, and A is the gain factor for the activation phase of phototransduction (Lamb and Pugh, 1992; Pugh and Lamb, 1993). The number of photoisomerizations, Φ, is the product of the stimulus strength, i (photons per square micrometer at λmax), andA c, the effective collecting area of the outer segment, calculated as by Baylor et al. (1984): Equation 2where V OS is the volume of the outer segment; Q isom is the quantum efficiency of photoisomerization (0.67)(Dartnall, 1972);f is a factor allowing for the use of unpolarized light entering the outer segment perpendicular to its long axis (f = 0.5); and α is the specific pigment density (0.016 μm−1) (Bowmaker et al., 1980). Typically stimuli of the optimum wavelength were used; if not, the stimulus strengths were converted to the equivalent number of photons at the optimum wavelength, based on the spectral sensitivity function (see Table 2). The signals from rods were digitized at 4 msec intervals and typically low-pass-filtered at 50 Hz. The signals from cones were digitized at 3 msec intervals and typically low-pass-filtered at 100 Hz.
For seven cells, individual outer segment volume was calculated from digital images of the cells taken during or after recording. The average volume for rods (n = 4) was 11.9 μm3 and for cones (n = 3) was 14.3 μm3, corresponding to collecting areas of 0.147 and 0.175 μm−2 respectively. For nine other cells (five rods and four cones), the outer segment volumes and collecting areas were assumed to be similar to the measured means. The small size of mammalian outer segment dimensions and the resolution of the light microscope limit these volume estimates to an accuracy of ∼20–30%, based on a linear measurement error of 0.2 μm. The outer segment of cone photoreceptors is physically fragile but physiologically sturdy; although only a small stub of the OS remained in some cases, high-quality stable recordings were obtained.
For each cell, two to five rising phase responses were simultaneously fit by Equation 1, where A andt d were allowed to vary to optimize the fitting (Igor; Wavemetrics Inc.).R max was fixed as the value measured in each cell.
RNA isolation, reverse transcription-PCR, and cDNA cloning.Total RNA was extracted from chipmunk whole retina with retina pigment epithelium using Trizol reagent (Invitrogen, San Diego, CA) by following manufacturer's instructions. Chipmunk RGS9-1, Gβ5S, and cone PDEγ cDNAs were cloned by reverse transcription (RT)-PCR and rapid amplification of cDNA ends (RACE) strategies as described previously (Davis et al., 1994). A cDNA fragment encoding amino acids 327–394 within the conserved RGS domain of RGS9-1 was amplified by RT-PCR using degenerate primers cRGS9a, 5′-GGNTT(C/T)TGGGA(A/G)GCNTG(C/T)GAGA-3′; and cRGS9b, 5′-CAT(A/G)TA(A/G/T)AT(A/G)TGNGT(C/T)TGNGCNGC(A/G)TC-3′. To obtain the coding sequence on the 5′ end of this fragment, PCR was performed using a degenerate primer, cR5UTR, 5′-T(C/G/T)(A/C)(A/G)TCCAGG(A/G)(G/T)CCAG-3′, corresponding to the conserved sequence within the 5′ untranslated regions (UTRs) of RGS9-1 from different species, and primer cRNon, 5′-GTAGCGGTGGGGGTG-3′, which is reverse complementary to nucleotides encoding amino acids 379–383. The rest of the coding sequence was amplified by RT-PCR using primer cRCod, 5′-CACGGTGAAGGGGCTGAAG-3′, encoding amino acids 373–378; and degenerate primer cRCnon, 5′-TTA(C/T)TTNGGNGGNAG(C/T)TC(C/T)TT-3′, designed to be the reverse complement of nucleotides encoding amino acids 479 to stop code, assuming conservation of the last six amino acids with bovine, human, and mouse sequences. A fragment of chipmunk Gβ5S cDNA encoding the first 338 amino acids was cloned by RT-PCR using degenerate primers cGa, 5′-ATGGCNACNGANGGN(C/T)T-3′; and cGb, 5′AANACNGTNCC(A/G)TCNGG3′. To obtain upstream sequence, RT-PCR was performed using degenerate primer cG5UTR, 5′-CCG(C/G)(A/G)CGAAGATGGC-3′, which is conserved in 5′ UTRs of Gβ5S; and primer cGNon, 5′-GGTCTTCATGACAAACTG-3′. The rest of the coding sequence was obtained by 3′ RACE using primers cG3raceI, 5′-GAGTCTCCATCCTGTTTG-3′; cG3raceII, 5′-GTACTCTACGAGTCTC-3′; and adaptor, 5′-GACTCGAGTCGACATCG-3′. Chipmunk cone PDEγ cDNA was cloned by RT-PCR using degenerate primers cP5UTR, 5′-GCCG(A/C)CC(A/G)GGGG(A/C)AGT(C/T)AAAATG-3′; and cP3UTR, 5′-TGGCAGAACC(C/T)CTGG(C/T)(A/G)CT-3′. These sequences are conserved in 5′ and 3′ UTRs of mammalian cone PDEγ.
Sequence data analyses. The chipmunk RGS9-1, Gβ5S, and cone-type PDEγ (GenBank accession numbers AF480878, AF480879, and AF480880, respectively) were compared with the corresponding sequences, which were taken from the GenBank or National Center for Biotechnology Information database, with the following accession numbers: mouse (Mus musculus) RGS9-1,AAC99481; Gβ5, P54314; cone PDEγ, BAB32255.1; and rod PDEγ, CAA68714.1; rat (Rattus norveqicus) cone PDEγ, AAG43400.1; human (Homo sapiens) RGS9-1, AAG09311; Gβ5, AAC63826; cone PDEγ, BAA08241.1; and rod PDEγ, AAA03653.1; bovine (Bos taurus) RGS9-1, O46469; cone PDEγ, AAA30689.1; and rod PDEγ, CAA28507.1; tiger salamander (Ambystoma tigrinum) Gβ5,AAK52836.1; fruit fly (Drosophila melanogaster) Gβ5, AAF46336; nematode (Caenorhabditis elegans) Gβ5, AC Q20636; 13-lined ground squirrel (Spermophilus tridecemlineatus) cone PDEγ,CAA04720.1; leopard frog (Rana pipiens) cone PDEγ,AAK95403.1; and rod PDEγ, AAK95404.1; guinea pig (Cavia porcellus) rod PDEγ, AAG43274.1; and dog (Canis familiaris) rod PDEγ, CAA93815.1. The sequences were aligned, and phylogenetic trees were constructed by CLUSTALW (Thompson et al., 1994) based on the neighbor-joining method (Saitou and Nei, 1987).
Protein expression and purification. Chipmunk RGS9-1 and PDEγ and mouse RGS9-1 cDNAs were subcloned into the pET-14b expression vector (Novagen, Madison, WI) usingNdeI and BamHI restriction sites. His6-tagged recombinant proteins were purified using a Ni+-nitrilo triacetic acid column (Qiagen, Hilden, Germany) by following the manufacturer's instructions using denaturing conditions. His6-tagged chipmunk PDEγ was further purified by reverse-phase HPLC as described previously (Angleson and Wensel, 1994). Endogenous bovine rod PDEγ was purified from the bovine rod outer segment (ROS) following procedures described previously (Wensel and Stryer, 1990). In each case, the concentration of purified protein was determined by absorbance at 280 nm in 6 mguanidinium chloride using extinction coefficients calculated from the sequence (Gill and von Hippel, 1989).
Immunofluorescence staining. Immunofluorescence staining of mouse and chipmunk RGS9-1 and Gβ5 was performed according to the procedure described previously (Lyubarsky et al., 2001). To stain chipmunk RGS9-1 and rhodopsin, chipmunk eyes were fixed in 4% paraformaldehyde and PBS, pH 7.2, for 10–16 hr at 4°C. After protection in 30% sucrose and PBS for 1 hr at 4°C, the eyes were embedded in OCT compound (Tissue-Tek), and frozen sections were cut at 16 μm. Tissue sections were postfixed in 1:1 methanol/acetone (v/v) for 10 min at room temperature and rehydrated in PBS, pH 7.2, for 20 min at room temperature. Nonspecific binding was blocked by incubating the sections for 1 hr at room temperature with 10% sheep serum (Sigma, St. Louis, MO) and PBS. Then the sections were incubated with primary antibody to RGS9, anti-RGS9-1c (He et al., 1998), at a 1:200 dilution, and anti-rhodopsin monoclonal antibody, 1D4 (Wu et al., 1998), at a 1:500 dilution in 10% sheep serum and PBS. The sections were incubated with primary antibodies at room temperature overnight in a humidified atmosphere. After being washed three times for 5 min in PBS at room temperature, sections were incubated with secondary antibodies, fluorescein isothiocyanate-conjugated anti-rabbit IgG (Vector Laboratories, Burlingame, CA) and rhodamine-conjugated anti-mouse IgG (Vector Laboratories), at a 1:100 dilution, in 10% sheep serum and PBS for 1 hr at room temperature in a humidified atmosphere. Sections were washed three times for 10 min in PBS at room temperature and mounted in aqueous mounting medium (Gel/Mount; Biomeda, Foster City, CA). Sections were examined and images recorded using a Zeiss (Thornwood, NY) 510 LSM confocal microscope.
Preparation of retina extracts. All procedures were performed in complete darkness or under infrared light. Because of the extracellular matrix structures known as cone sheaths, chipmunk retina is not easily peeled from the underlying layer of retinal pigmented epithelium (RPE) without substantial loss of cone outer segments. Also, previous studies have shown that the proteins of interest are not expressed at detectable levels in RPE; therefore, we homogenized a sample of retina plus RPE to maximize the yield of cone-derived material. One chipmunk retina with attached retinal pigmented epithelium, two mouse retinas, and 100 μg of bovine retina were homogenized in 600 μl of buffer A. After centrifugation for 15 min at 80,000 × g, the pellets were resuspended in 300 μl of buffer A supplemented with 3 μl of 4 mmethanol solution of 11-cis-retinal and incubated at 4°C for 2 hr. Retina pellets were recovered by centrifugation again for 15 min at 80,000 × g and incubated in buffer A supplemented with 1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate for 30 min at 4°C. Supernatants containing extracted membrane and soluble proteins were obtained by centrifugation for 15 min at 10,000 ×g.
Spectroscopic measurements. The UV-visible absorption spectra of detergent-solubilized retina extracts were recorded with an HP 8452A spectrophotometer in cuvettes of 1.0 cm path length at room temperature. Spectra were recorded before and after illumination of samples for 3 min with a 150 W light bulb (Reflector Flood; Philips). Hydroxylamine-hydrochloride, pH 7.0, was added to a final concentration of 20 mm after the samples were illuminated, and the spectra were recorded again. Difference spectra were obtained by subtracting the spectrum after bleaching from that before illumination. Absorbance values were all within the linear response range of the spectrophotometer. To generate the model spectra for rods and M-cones, the Dawis (1981) polynomials were used in a two-parameter least squares fit varying the log of maximum sensitivity,b max, and peak wavelength, λmax. To generate estimates of extinction coefficients, the sensitivity values at each wavelength were normalized by dividing by 10bmax and then multiplied by 40,000m −1 · cm−1(Vissers et al., 1998). For the S-cone sensitivity data and for the M-cone data for wavelengths <450 nm, in which the Dawis polynomial fits poorly, the data were fit to Gaussian curves,S(λ) = a o + (S(λc) −a o)exp[−(λ −λc)2/2w 2], to provide a smooth curve declining monotonically with distance from λc and closely approximating the sensitivity data (Ditchburn, 1963; Harris and Bertolucci, 1978), and the extinction coefficients were calculated as ε(λ) = S(λ) · 40,000 M−1 · cm−1/S(λc). The assumptions are that the extinction coefficients ε(λ) are proportional to the sensitivities S(λ) with a constant proportionality constant for each cell type and pigment, and that the extinction coefficient ε(λmax or λc) at the wavelength of maximal sensitivity (λmax or λc) is equal to 40,000 M−1 · cm−1. Values found by fitting were as follows: rods,L max = 502 nm; λmax = 500.36 nm; andb max = 0.0058 (b max = 0 assigned to maximal measured value tabulated in Table 2); M-cones,L max = 562 nm; λmax = 537.48 nm;b max = 0.0066;a o = 0.21384;S(λc) = 1.126; λc = 538.78 nm; and w = 44.34 nm; and S-cone, a o = −0.0206;S(λc) = 1.324; λc = 452.8 nm; and w = 58.49 nm. The difference spectrum was fit to a linear combination of the model spectra in the range of 400–600 nm using the Levenberg–Marquardt (Levenberg, 1944; Marquardt, 1963) least squares algorithm as implemented in the program Origin and the equation ΔA(λ) = mεm(λ) + rεr(λ) +sεs(λ) − (m +r + s)εR∗(λ). Herem, r, and s are the molar concentrations of M-cone pigment, rhodopsin, and S-cone pigment, respectively; εm(λ), εr(λ), and εs(λ) are the corresponding model spectra from Figure 3 A; and εR∗(λ) is the metarhodopsin II spectrum shown in Figure 3 A.
Immunoblotting and densitometry. After spectrophotometry to quantify visual pigments, chipmunk, mouse, and bovine retina detergent extracts and bovine ROS were analyzed by SDS-PAGE, followed by immunoblotting and densitometry. Immunoblotting was performed according to a standard protocol (Harlow and Lane, 1988) on proteins separated by SDS-PAGE. Buffer B was used for electrophoretic transfer of PDEγ, and buffer B supplemented with 0.1% SDS was used for transfer of RGS9-1 and Gβ5. The membranes for immunoblotting were supported nitrocellulose (NitroPure; Osmonics, Inc.). After 60 min for RGS9-1 and Gβ5 or 45 min for PDEγ transfer at 350 mA at 4°C, membranes were blocked by 5% nonfat dry milk and a solution of 20 mm Tris-HCl, pH 7.2, 150 mm NaCl, and 0.1% (v/v) Tween 20 for 1 hr, followed by incubation with primary antibody for 4 hr. Polyclonal antibodies anti-RGS9-1c and anti-Gβ5 (He et al., 2000) were used at a 1:1000 dilution, and anti-PDEγ was used at a dilution of 1:500. The secondary antibody used was horseradish peroxidase-conjugated anti-rabbit IgG (Promega, Madison, WI), with detection by chemiluminescence using the ECL system (Amersham Biosciences, Arlington Heights, IL). For densitometry of chemiluminescence signals on film, x-ray films were scanned, and bands were quantified by software UN-SCAN-IT (Silk Scientific Corp.). To quantify RGS9-1 and PDEγ or to calibrate antibody specificity, purified recombinant proteins and highly purified PDEγ from bovine retina were used as standards. The concentrations of purified proteins were determined by spectrophotometry at 280 nm as described above or by using densitometry of Coomassie blue-stained bands on SDS-PAGE gels calibrated with standards whose concentrations were determined by 280 nm absorbance. On each gel, varying amounts of standard protein were loaded next to different volumes of the retinal extract. Films were exposed to the blots after processing for chemiluminescence detection, with varying exposure times to ensure that film optical density was linear with the protein amount for standards whose values of optical density bracketed those of the corresponding bands from the extracts. The optical density of the band in question from each extract lane was then substituted into the linear function (derived by least squares fitting) for density, to solve for the amount of specific protein in each extract sample. Dividing this amount by the volume loaded gave a value of concentration for each sample, and these were averaged. In some cases, the blot was initially performed with a standard from another species (e.g., bovine PDEγ), and then later, the relative sensitivity of the antibody for the protein of the species in question (e.g., chipmunk) was determined using blots with purified recombinant proteins from both species (e.g., bovine and chipmunk) to obtain correction factors. The correction factors were determined as the ratios of calculated amounts of protein from Western blots and densitometry to the amounts of each purified protein (determined by spectrophotometry) loaded on SDS-PAGE (see Figs. 6 B,7 C,D). The correction factor for chipmunk cone PDEγ detection by anti-PDEγ is 0.79. The correction factors of anti-RGS9-1c for chipmunk and mouse RGS9-1 are 0.97 and 1.97, respectively.
Immunoprecipitation. Purified anti-RGS9 antibody anti-RGS9-1c was covalently attached to cyanogen bromide-activated Sepharose 4B-CL as described previously (Hu et al., 2001). For immunoprecipitation, chipmunk or bovine retina was homogenized and solubilized with 200 μl of buffer A supplemented with 1% Nonidet P-40. The insoluble material was removed by centrifugation for 15 min at 80,000 × g. The solubilized retina extracts were incubated with 10 μl of anti-RGS9-1c IgG-coupled beads for 10–16 hr at 4°C after mixing on a shaker. The beads were separated from supernatant by a brief centrifugation and washed three times with the solubilization buffer. Bound proteins were redissolved in the SDS-PAGE sample buffer and separated from the beads by a brief centrifugation.
PDE assays. PDE catalytic activity was measured with the pH-recording method (Liebman and Evanczuk, 1982) as modified previously (Malinski and Wensel, 1992). Specifically, assays were performed in buffer C with initial cGMP concentration of 2 mm and a total volume of 200 μl. Assays were performed in 96-well microtiter plates and monitored with MI-410 microelectrodes (Microelectrodes, Inc.). To test chipmunk cone PDE and bovine rod PDE activity, each assay was initiated by adding 15 μl of chipmunk retina homogenate or 10 μl of bovine ROS, which had 0.4 pmol of PDE calculated from PDEγ quantitative immunoblots, and maximal PDE activation was obtained by adding 30 μg of trypsin (Sigma) to remove inhibitory subunit PDEγ. After PDE was fully activated, 300 μg of soybean trypsin inhibitor (Sigma) was added to quench trypsin. Then PDE activity was blocked by adding 0.2 nmol of His6-tagged bovine PDEγ and restored by adding 300 μg of trypsin again. To test chipmunk cone PDEγ and bovine rod PDEγ inhibitory activity, bovine PDE was purified and treated with trypsin as described previously (Wensel and Stryer, 1990). Each assay was performed in 200 μl of buffer C with 2 mm cGMP, and the pH recordings were initiated when trypsin-treated PDE was added to a final concentration of 5 nm. After the reactions had proceeded ∼1 min, His6-tagged bovine rod or chipmunk cone PDEγ was added to different concentrations of 0, 2, 4, and 8 nm. To test chipmunk cone PDEγ inhibitory activity on chipmunk PDE, one chipmunk retina was homogenized in 200 μl of buffer C, and 150 μl of the homogenate was treated by 150 μg of trypsin at room temperature for 1 min, followed by 1.5 mg of soybean trypsin inhibitor. Each pH assay was performed in a final volume of 200 μl of buffer C with 2 mm cGMP. The recordings started when 40 μl of trypsin-treated chipmunk retina homogenate (∼0.4 pmol of PDE) was added. After the reactions had proceeded ∼1 min, His6-tagged chipmunk cone PDEγ was added to different concentrations of 1, 2, and 4 nm.
GTPase single turnover assays. GTPase single turnover assays were performed to test the bovine rod or chipmunk cone PDEγ RGS9-1 GTPase-accelerating protein (GAP) enhancement effect, essentially as described previously (He et al., 2000). Specifically, bovine rod outer segments containing 10 μm rhodopsin were exposed to light and mixed with different amounts (0, 10, 20, 50, or 100 nm) of His6-tagged bovine rod or His6-tagged chipmunk cone PDEγ in buffer D. Then GTP hydrolysis was initiated by adding 7 μl of [γ-32P]GTP (Amersham Biosciences) to 14 μl of the above mixture by vortexing. The reaction was quenched by 100 μl of 5% trichloroacetic acid at various times, and Pi (free phosphate ion) released from hydrolyzed GTP was determined by activated charcoal assay. The first-order rate constants for GTP hydrolysis (k inact) were obtained by fitting data to single exponentials.
Immunostaining of RGS9-1 and Gβ5 in mouse and chipmunk retinas
RGS9-1 immunofluorescence was observed primarily in outer segments of photoreceptor cells (Fig.1 A,B,D,E). In rod-dominant mouse retina, much brighter immunofluorescence was observed in cone outer segments, identified by staining of cone sheaths with rhodamine-conjugated peanut agglutinin. In cone-dominant chipmunk retina, cone outer segments were stained brightly, whereas RGS9-1 staining in rod outer segments, identified by staining with anti-rhodopsin antibody 1D4, was barely detectable. These results, together with previous results in bovine and human retinas (Cowan et al., 1998; Zhang et al., 1999) confirm that in mammals, much more RGS9-1 is present in cone outer segments than in rods, and that most RGS9-1 in chipmunk retina is in cone outer segments.
Strong Gβ5 staining (Fig.1 C,F) was observed in photoreceptor outer segments. The Gβ5 staining of the multiple chipmunk rods present in each field was so weak as to be undetectable, whereas the cones stained brightly (Fig. 1 F). There is also staining in what appear to be the photoreceptor synaptic termini in the outer plexiform layer. The function of Gβ5 in this region of the cells remains to be determined. Because RGS9-1 staining is not observed in the outer plexiform layer, any Gβ5 there is unlikely to be associated with RGS9-1.
Photoresponses of chipmunk cones
Suction electrode recordings were made from chipmunk rod and cone cells to define the differences in their sensitivity and the kinetics of the recovery phase of their light response as well as to determine the action spectra. Figure 2 shows families of photocurrent responses in a rod (Fig. 2 A) and M-cone (Fig. 2 B) to brief light flashes of increasing intensity. The average time to peak of the linear range responses in the rods was 2.25 times slower than in M-cones (Table1). The graphs on theright of Figure 2, A and B, plot the peak response amplitude versus the log-stimulus intensity. Thesmooth curves are fits to the data with an exponential saturation function of the type used by Lamb et al. (1981). There was a 38-fold difference in the sensitivity as measured byI 1/2 (the stimulus strength required to produce a half-maximal response) or 140-fold difference as measured by the flash sensitivity (S f,picoamperes per photon per square micrometer), which is measured for linear range responses (Table 1).
Use of RGS9-1 staining to identify cones and 1D4 staining to identify rods allowed counting of rods and cones in the chipmunk sections. In eight fields counted, cones made up an average of 75.0 ± 2.6% of the total photoreceptors; if we account for the slightly higher probability of finding cones in an optical section because of their ∼11% greater width, this value becomes 68%. Of the 15 cones studied electrophysiologically, two were S-cones; the remainder were M-cones. Eleven of 13 M-cones and 0 of 2 S-cones had an undershoot in the recovery phase of the response.
The efficiencies of signal amplification in the transduction cascade in different cell types can be compared by fitting the rising phases of the responses to Equation 1, which models the activation reactions only, and then comparing the values of A, or gain. Results of the model fitting to multiple responses are shown for one rod (Fig.2 C) and one cone (Fig. 2 D). The mean values for 16 chipmunk rod and cone cells are given in Table 1. The gain for cones had a mean value of 1.0 sec−2, and a range of 0.3–1.7 sec−2. Similar results were obtained when the same analysis was applied to previously published results of human cone (Kraft et al., 1998, their Fig. 1; A = 0.9 sec−2) and ground squirrel cone (Kraft, 1988, Fig. 3 a,b;A = 1.0 and 1.7 sec−2, respectively). Chipmunk rods had a mean gain of 10.4 sec−2 with a range of 6.2–14.2 sec−2. Although there is some variability from cell to cell for both rods and cones, the average values indicate an order of magnitude lower amplification, as reflected in the parameter A, in cones of all three species compared with chipmunk rods.
To achieve some appreciation for the difference in the recovery kinetics of the photoresponses, a simple linear regression of the photocurrent recovery was performed covering saturating and semisaturating responses. The recovery kinetics were also significantly faster in cones than in rods. As indicated in Table 1, the slopes of the recovery phases (picoamperes per second) for the cones were 3.1- to 4.5-fold steeper than those of rods, with both normalized to the maximum photocurrent in each cell.
The action spectra of the rod and cone visual pigments were measured by determining the stimulus strength that was required to generate a criterion linear range response. The results are presented in Table2 and graphically in Fig. 3 A. S- and M-cones were encountered with peak sensitivities of ∼450 and 540 nm, respectively. The rod peak sensitivity was 504 nm; no L-cones were encountered. Behavioral testing of the cone-mediated vision found the threshold difference between 539 and 580 nm of 0.231 log units (R. E. Van Arsdel and M. S. Loop, personal communication), an excellent match of the action spectra sensitivity differences for the M-cones from single-cell recordings at similar wavelengths (0.239 log units for 541 vs 580 nm; Table 2).
Cloning of chipmunk RGS9-1, Gβ5, and cone PDEγ cDNAs
Chipmunk RGS9-1, Gβ5S, and PDEγ cDNAs were cloned from chipmunk retina via RT-PCR and RACE; sequences are available from GenBank with accession numbers listed in Materials and Methods. The cloned chipmunk RGS9-1 cDNA includes part of the 5′ UTR and coding sequence corresponding to first 478 of 484 amino acids. The rest of the coding sequence was cloned by degenerate PCR on the basis of the assumption that the last six amino acids are conserved among mammalian RGS9-1. The deduced amino acid sequence of chipmunk RGS9-1 shares >90% identity with human, mouse, and bovine RGS9-1 (Fig. 4 A). By RT-PCR, chipmunk cone PDEγ cDNA was cloned, using two degenerate primers conserved in the 5′ and 3′ UTR of mammalian cone PDEγ. The deduced amino acid sequence of chipmunk PDEγ has 95% identity to bovine, mouse, ground squirrel, and human cone PDEγ and 78% identity to bovine, human, mouse, and dog rod PDEγ (Fig. 4 C). Chipmunk Gβ5S cDNA was also cloned from chipmunk retina RNA by RT-PCR and RACE. The deduced amino acid sequence is 100% identical to that of mouse Gβ5S (Fig.4 B).
Catalytic activity of chipmunk cone PDE
To measure the efficiencies of hydrolysis of cGMP by PDE in mammalian cones and compare the results with those in rods, we performed pH-based PDE assays in homogenates from cone-dominant chipmunk retina and bovine ROS. Figure5 A shows pH recordings in bovine or chipmunk samples containing 4 nmendogenous PDEγ. Both chipmunk cone and bovine rod PDE displayed low basal activity and were similarly activated by addition of trypsin to remove the inhibitory subunit PDEγ, and the activation was reversed by adding back His6-tagged bovine rod PDEγ after the addition of soybean trypsin inhibitor. Basal and maximal PDE activities in chipmunk retinal homogenates were similar to those in bovine samples. The concentration of PDE in each of the samples was calculated to be 2 nm from PDEγ immunoblots, using the ratio of two PDEγ subunits per holoPDE heterotetramer. On the basis of this concentration and the observed catalytic activity, the calculated maximum turnover numbers for chipmunk and bovine PDE were 3600 and 3900 mol of cGMP hydrolyzed per mole of PDE per second, respectively. These numbers are consistent with a previously reported rod PDE k cat value of ∼4000 cGMP hydrolyzed per PDE per second (Hurley and Stryer, 1982; Stryer et al., 1983), indicating chipmunk cone PDE and bovine rod PDE have similar cGMP hydrolytic activities when fully activated. Bovine cone PDE has been reported to have a similar specific activity of 3500–4670 cGMP per second (Gillespie and Beavo, 1988).
Biochemical properties of chipmunk cone PDEγ
To test the potency of PDE inhibition by chipmunk cone PDEγ, pH assays were performed using 5 nm trypsin-treated bovine PDE or 2 nm chipmunk PDE, and varying concentrations of bovine (0, 2, 4, and 8 nm) or chipmunk (0, 1, 2, 4, and 8 nm) His6-PDEγ (Fig. 5 B). Hydrolysis of cGMP slowed dramatically within seconds of addition of recombinant PDEγ. cGMP hydrolytic velocity,d[cGMP]/dt, was determined from the slope at each point along the pH-recording traces (Fig. 5 B,top). PDEγ inhibitory activity toward trypsin-activated PDE was determined by the difference in hydrolytic velocity before and after addition of PDEγ. Figure 5 B, bottom, shows that chipmunk cone and bovine rod PDEγ have similar inhibitory effects on bovine rod PDE. The PDEγ inhibitory activities were calculated from linear least squares fits. When added to bovine rod PDE, each mole of bovine rod PDEγ inhibited hydrolysis of 1773 ± 207 (mean ± SD) mol of cGMP per second, and each mole of chipmunk cone PDEγ inhibited hydrolysis of 1620 ± 275 mol of cGMP per second. When added to chipmunk PDE, each mole of recombinant chipmunk cone PDEγ inhibited 1781 ± 109 mol of cGMP hydrolysis per second, indicating similar potencies in PDE inhibition, as observed previously for bovine cone PDEγ (Hamilton et al., 1993).
Rod PDEγ can enhance the GAP activity of RGS9-1/Gβ5L (He et al., 2000; Skiba et al., 2000). To find out whether chipmunk cone PDEγ has similar RGS9-1 GAP activity enhancement activity, GTPase single turnover assays were performed in bovine ROS containing 10 μm rhodopsin (Fig.5 C). Chipmunk cone and bovine rod recombinant PDEγ increased RGS9-1/Gβ5L-mediated hydrolysis of GTP by transducin with similar potencies; the same concentrations of recombinant proteins led to similar GTP hydrolysis rates, with a threefold acceleration above basal hydrolysis at a concentration of 50 nm in each case.
Quantification of proteins
To quantify the levels of phototransduction proteins that determine the lifetime of effector activation through regulation of the kinetics of G-protein GTP hydrolysis, we determined the molar ratios of RGS9-1 and PDEγ to visual pigments in cone-dominant chipmunk and rod-dominant mouse and bovine retinas. The amounts of visual pigments were quantified by difference spectrophotometry. Figure 3 Bshows the chipmunk retina–RPE extract difference spectrum, with a maximal value at 534 nm. Because in chipmunk retina, there are three kinds of visual pigments, rhodopsin, mid-wavelength-sensitive (M-cone) pigment, and short-wavelength-sensitive (S-cone) pigment, the difference spectrum is a linear combination of the difference spectra of the two cone pigments and rhodopsin. To quantify cone pigments, we generated model spectra (Fig. 3 A) on the basis of our spectral sensitivity data (Table 2) as described in Materials and Methods and used a measured spectrum of bovine metarhodopsin II as an approximation of the spectra for all three photoexcited chipmunk pigments (Fig. 3 A). Figure 3 B shows a linear combination of the model spectra from Figure 3 A fit to the observed difference spectrum in the range of 400–600 nm. Thesmooth curve is the result predicted for a molar ratio of M pigment/rhodopsin/S pigment of 1:0.3:0.05, assuming equal extinction coefficients at the absorbance peak for each. The proportion of total visual pigment contributed in this fit by rhodopsin, 22%, is consistent with our immunofluorescence results indicating that rods make up 25% of total photoreceptors. Our results do not provide an accurate estimate of the relative numbers of S-cones or the amounts of S pigments, although they are clearly much lower than the numbers for M-cones and rods. By using the reasonable assumption that the cone visual pigments' extinction coefficients are similar to those of other visual pigments (∼40,000m −1 · cm−1;Vissers et al., 1998), a total cone pigment concentration of 0.60 μm was determined for the sample shown. The amounts of rhodopsin in rod-dominant bovine and mouse retina and bovine ROS samples were quantified by difference spectrophotometry as well.
To determine the amounts of RGS9-1 and PDEγ from different samples, Western blots, followed by densitometry, were performed, and the results were compared with standard curves generated with purified proteins as described in Materials and Methods.
Similar molar ratios of PDEγ to visual pigments were obtained in chipmunk, 1:68, mouse, 1:65, and bovine retinas, 1:76 (Fig.6 D), indicating similar relative concentrations of PDE in cones and rods. Bovine rod outer segments, purified from frozen retinas, had a slightly lower ratio of PDEγ to rhodopsin, 1:104, likely because of loss of the soluble form of PDE associated with the PDEδ subunit (Gillespie et al., 1989;Florio et al., 1996). The molar ratio of RGS9-1 to cone pigments in chipmunk retina was determined (Fig.7 E) to be ∼1:62, >14-fold higher than that in purified bovine ROS, 1:910, and ∼10 times higher than the ratio in bovine and mouse retina, 1:610. Because the molar ratios of visual pigments to transducin are not very different between rods and cones (Tachibanaki et al., 2001), these results indicate that there is a 10-fold higher ratio of RGS9-1 to transducin in cones than in rods. The total concentrations of RGS9-1 and PDEγ, which work together to achieve maximal transducin GTPase acceleration, are very similar in cones, whereas in rods there is an almost 10-fold excess of PDEγ over RGS9-1. The higher concentration of RGS9-1 in cones may be important in the faster recovery of light responses in cones.
Both Gβ5L and Gβ5S bind to RGS9-1 in chipmunk cones
RGS9-1 forms a complex with Gβ5L in photoreceptor outer segments of rod-dominant mouse and bovine retinas, and the proteins are mutually dependent on one another for expression and stability (Makino et al., 1999; Chen et al., 2000; He et al., 2000). From previous studies, the molar ratio of RGS9-1 to Gβ5L appears to be very close to 1:1 in rod photoreceptors. To find out whether the molar ratio of RGS9-1 to Gβ5L is similar in cone outer segments, we performed immunoblots of chipmunk retina using Gβ5 antibodies. Surprisingly, a much smaller ratio of Gβ5L to RGS9-1 immunoblot signal was found in chipmunk retina homogenate compared with that in bovine retina homogenate (Fig. 8) on the same blot. The difference in immunoblot Gβ5L signal is unlikely to be attributable to differences in antibody sensitivity, because the epitope recognized by our antipeptide antibodies is identical in bovine and chipmunk Gβ5L. To find out whether some RGS9-1 is associated with the short form, Gβ5S, in chipmunk cones, RGS9-1 GAP complexes were immunoprecipitated from chipmunk retina homogenate using anti-RGS9-1c antibody. In the precipitated pellets, Gβ5L and Gβ5S were found in similar amounts (Fig. 8). Because the measurements described above indicate that >96% of the RGS9-1 in chipmunk retina is in cones, this result indicates that, in contrast to the exclusive association of RGS9-1 with Gβ5L in bovine and murine rods, RGS9-1 associates with similar amounts of Gβ5L and Gβ5S in chipmunk cones.
The differences observed in rod and cone photoresponses in the chipmunk, like differences between rods and cones of other species, likely arise at several steps in the phototransduction cascade. Candidates for key biochemical differences giving rise to lower sensitivity and faster recovery are the photopigments themselves, lifetimes of activated pigments, activation of the G-proteins by light-activated pigments, lifetimes of activated G-proteins, G-protein–effector coupling, cGMP-gated channel properties, and differences in Ca2+ concentrations and Ca2+-regulatory mechanisms feeding back on the other steps and on guanylate cyclase activity. There have been only a few biochemical studies of isolated cone pigments, but these suggest that the complex with 11-cis-retinal may be less stable than that formed by rhodopsin both with respect to thermal isomerization (Birge and Barlow, 1995), leading to greater dark noise, and with respect to dissociation of retinal, which could lead both to increased background attributable to weak activation by apo-opsins on dissociation of 11-cis-retinal in the dark and to rapid inactivation attributable to rapid dissociation of all-transretinal after photoactivation. A recent study of isolated carp cones (Tachibanaki et al., 2001) reported that the gain of phototransduction reactions was much lower than in rod outer segments; however, direct measurements of G-protein activation by human and chicken green-sensitive cone pigments (Imai et al., 1997; Vissers et al., 1998) and Xenopus short-wavelength visual pigments (Starace and Knox, 1997; Babu et al., 2001) suggest that the efficiency of rod transducin activation by cone pigment equivalents of metarhodopsin II is only approximately twofold lower than activation by rod metarhodopsin II. Thermal decay of the metarhodopsin-like species has been consistently reported to be much faster for cone pigments than for rhodopsin, and the study of carp cones also revealed much faster phosphorylation and phosphorylation-induced inactivation. Thus the catalytic efficiency of photoexcited cone pigments is less likely to account for the differences in sensitivity than is the dramatically reduced lifetime of this catalytically active state. Rapid truncation of the rise in activated transducin is consistent with our observation of more rapid times to peak in chipmunk cone responses than in rod responses and with the 10-fold lower effective amplification.
Rapid inactivation of photoexcited pigment by itself is not sufficient to account for the rapid recovery kinetics observed in cones of chipmunks and other species. Recovery requires both restoration of cGMP levels through guanylate cyclase activation by light-induced reduction in intracellular [Ca2+] and recovery of phosphodiesterase activity to its dark state. Results from RGS9 knock-out mice (Chen et al., 2000; Lyubarsky et al., 2001) have revealed that prompt deactivation of phosphodiesterase depends in both rods and cones on RGS9-1-mediated acceleration of G-protein GTP hydrolysis. It is striking that although the identical GAP protein is present in rods and cones, their concentrations differ by more than an order of magnitude. Although genetic deletion of RGS9-1 did not measurably affect the rising phases or amplitudes of rod responses (Chen et al., 2000), because acceleration of transducin GTP hydrolysis is likely to occur much more rapidly in cones as a result of high RGS9-1 concentrations, GTP hydrolysis may contribute in cones to lowering gain in the rising phases of the responses as well. The instantaneous rate of production of activated G-protein in this phase is the difference between the rate of activation by light-activated pigment and the rate of inactivation by GTP hydrolysis. Thus rapid GTP hydrolysis in cones may contribute somewhat to lowered sensitivity as well as to faster recovery. If so, slowing of GTP hydrolysis by inactivation of RGS9-1 would be expected to increase sensitivity. Electroretinogram recordings of cone responses in mice lacking a functional RGS9 gene do indeed show a more than twofold increase in sensitivity compared with wild-type mice (Lyubarsky et al., 2001). The sensitivity increase was detected in the cone b-wave, which is driven by cone responses but does not derive directly from them. Because the cone a-wave, which has a more direct contribution from the cones themselves, is a rather weak and noisy signal in mice, a sensitivity difference on the order of twofold cannot be reliably detected. The observation of a twofold difference in b-wave sensitivity and failure to detect a larger sensitivity increase in the a-wave both suggest that higher RGS9-1 concentrations in cones are not sufficient to explain the much lower gain and sensitivity of cones compared with rods that we have observed. It seems likely that RGS9-1 plays a major role in recovery kinetics and a lesser role in sensitivity.
An unexpected difference we observed between cones and rods is the presence of approximately equal proportions of each isoform of the GAP subunit Gβ5, bound to RGS9-1 in chipmunk cones, in contrast to the apparently exclusive association of the long isoform Gβ5L with RGS9-1 in rods (Makino et al., 1999;Chen et al., 2000). Mice lacking the RGS9 gene also lack the long variant Gβ5L but not the short variant Gβ5S (Chen et al., 2000), so it seems likely that the short variant is responsible for the synaptic staining observed; the antibodies used do not distinguish between the variants. Gβ5S can associate with RGS11, RGS6, or RGS7 (Posner et al., 1999; Kovoor et al., 2000; Witherow et al., 2000; Zhang and Simonds, 2000) and has been reported previously to be bound to RGS7 in retinal extracts (Cabrera et al., 1998). It is likely that one or more of these other RGS proteins is bound to Gβ5 in the outer plexiform layer. Two recent comparisons of recombinant proteins (He et al., 2000; Skiba et al., 2001) found little difference between the values of the physiologically relevant parameter for catalytic efficiency,k cat/K m, for full-length RGS9-1 complexed with Gβ5Lcompared with the complex with the short isoform Gβ5S. It remains to be determined why rod outer segments contain only one of the Gβ5 splice variants, whereas cone outer segments contain both.
In contrast to the striking differences in RGS9-1 levels, the enzymatic properties and concentrations of PDE appear to be very similar in rods and cones. Many of the differences in PDE sequences between rods and cones are conserved across a number of species, but few striking differences between the properties of rod and cone PDE variants have been detected. PDEγ has two different activities important for phototransduction, inhibition of PDE catalytic subunits and enhancement of RGS9-1 GTPase acceleration. These activities are also similar in the rod and cone variants. Although we were not able to assess the efficiency with which transducin activates PDE in cones in our crude retina preparations, bovine cone PDE has been reported previously to be activated by transducin more efficiently than rod PDE (Gillespie and Beavo, 1988).
Although our measurements suggest that GTP hydrolysis plays an important role in determining the differences in cone and rod photoresponses, whereas catalytic properties of PDE likely do not, there are clearly other biochemical differences underlying the differences in photoresponses. There is evidence suggesting that differences in regulation of and by Ca2+plays an especially important role in the differences between rods and cones. The fraction of photocurrent carried by Ca2+ in cones (∼35%) is substantially greater than that in rods (∼20%; Ohyama et al., 2000, 2002), whereas Ca2+ extrusion is faster in cones (Korenbrot, 1995; Sampath et al., 1999), so that flashes eliciting similar current changes will give rise to larger changes in [Ca2+] in cones than in rods. Moreover, the [cGMP] dependence of channel gating is sensitive to [Ca2+] over the physiological range in cones (Rebrik et al., 2000). However, the faster response kinetics of cones do not depend entirely on differences in Ca2+ homeostasis, because they are faster even when light-induced Ca2+ changes are prevented (Nakatani and Yau, 1989; Matthews et al., 1990).
A complete molecular explanation of the differences in rod and cone photoresponses will require further quantification of other key components of the transduction cascade, particularly those involved in Ca2+ feedback and pigment phosphorylation. From the results presented here, it seems clear that chipmunk and other cone-dominant animals present excellent substrates for such studies.
This work was supported by National Eye Institute Grants EY11900, EY07981, and EY10573 and by the Welch Foundation. We thank Derron Allen and Jerry Millican for technical assistance and Trevor Lamb for helpful discussions on fitting the phototransduction gain.
Correspondence should be addressed to Theodore G. Wensel, Department of Biochemistry and Molecular Biology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030. E-mail:.